Analysis overview

This systematic review and meta-analysis evaluated whether antipsychotic treatment reverses NMDA receptor antagonist–induced deficits in social preference in animal models. Effect sizes were calculated as Hedges’ g and synthesized using multilevel random-effects models to account for dependency between multiple outcomes within experiments and studies.

Study landscape and evidence distribution

Alluvial plot

Distribution of evidence across species, NMDA antagonists, and antipsychotics. Alluvial plot illustrating how effect sizes are distributed across animal species, NMDA receptor antagonists used to induce social deficits, and antipsychotic drugs tested for reversal.

Evidence maps

Evidence maps of experimental design characteristics.
Bubble size represents the number of effect sizes (k), and color indicates the mean Hedges’ g within each cell.

Main meta-analysis

Overall effect

## 
## Multivariate Meta-Analysis Model (k = 40; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  1.659  1.288     21     no         study_id 
## sigma^2.2  0.000  0.000     28     no  study_id/exp_id 
## 
## Test for Heterogeneity:
## Q(df = 39) = 353.192, p-val < .001
## 
## Number of estimates:   40
## Number of clusters:    21
## Estimates per cluster: 1-5 (mean: 1.90, median: 2)
## 
## Model Results:
## 
## estimate     se¹   tval¹     df¹   pval¹  ci.lb¹  ci.ub¹     
##    0.940  0.302   3.115   19.88   0.005   0.310   1.569   ** 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t-test and confidence interval, df: Satterthwaite approx)

Multilevel random-effects meta-analysis with robust variance estimation.

Orchard plot

Overall antipsychotic effect on reversing NMDA antagonist–induced social preference deficits. Orchard plot summarizing study-level pooled effects with multilevel heterogeneity.

Prediction interval for the overall effect

##    estimate     ci_lb    ci_ub     pi_lb    pi_ub
## 1 0.9397334 0.3101173 1.569349 -1.821002 3.700469

Prediction interval for the overall effect. The 95% prediction interval reflects expected variability in the true effect size of a future study beyond sampling error.

##            Component I.....
## 1           I2_Total   85.2
## 2        I2_study_id   85.2
## 3 I2_study_id/exp_id    0.0

Multilevel heterogeneity estimates (I²).

Publication bias

Funnel plots

Funnel plot using standard error.

Funnel plot using inverse square root of total sample size.

Precision Effect Test (PET)

## 
## Multivariate Meta-Analysis Model (k = 40; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  0.790  0.889     21     no         study_id 
## sigma^2.2  0.125  0.354     28     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 38) = 114.214, p-val < .001
## 
## Number of estimates:   40
## Number of clusters:    21
## Estimates per cluster: 1-5 (mean: 1.90, median: 2)
## 
## Test of Moderators (coefficient 2):¹
## F(df1 = 1, df2 = 3.88) = 94.901, p-val < .001
## 
## Model Results:
## 
##           estimate     se¹    tval¹    df¹   pval¹   ci.lb¹   ci.ub¹      
## intrcpt     -4.054  0.686   -5.913    5.5   0.001   -5.770   -2.338    ** 
## sqrt(vi)     8.514  0.874    9.742   3.88   <.001    6.058   10.970   *** 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

PET (Precision Effect Test) model with robust variance estimation. The PET model evaluates small-study bias by regressing effect size on study precision.

Precision Effect Estimate with Standard Error (PEESE)

## 
## Multivariate Meta-Analysis Model (k = 40; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  0.680  0.825     21     no         study_id 
## sigma^2.2  0.000  0.000     28     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 38) = 100.556, p-val < .001
## 
## Number of estimates:   40
## Number of clusters:    21
## Estimates per cluster: 1-5 (mean: 1.90, median: 2)
## 
## Test of Moderators (coefficient 2):¹
## F(df1 = 1, df2 = 2.88) = 36.592, p-val = 0.010
## 
## Model Results:
## 
##          estimate     se¹    tval¹     df¹   pval¹   ci.lb¹  ci.ub¹    
## intrcpt    -0.748  0.433   -1.727   12.15   0.110   -1.690   0.195     
## vi          4.678  0.773    6.049    2.88   0.010    2.156   7.201   * 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

PEESE (Precision Effect Estimate with Standard Error) model with robust variance estimation. The PEESE model provides an alternative bias-adjusted estimate using study variance.

Time-lag bias

## 
## Multivariate Meta-Analysis Model (k = 40; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  1.666  1.291     21     no         study_id 
## sigma^2.2  0.000  0.000     28     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 38) = 337.279, p-val < .001
## 
## Number of estimates:   40
## Number of clusters:    21
## Estimates per cluster: 1-5 (mean: 1.90, median: 2)
## 
## Test of Moderators (coefficient 2):¹
## F(df1 = 1, df2 = 7.03) = 2.130, p-val = 0.188
## 
## Model Results:
## 
##          estimate     se¹   tval¹     df¹   pval¹   ci.lb¹  ci.ub¹     
## intrcpt     0.916  0.292   3.133   18.17   0.006    0.302   1.530   ** 
## year_c      0.104  0.071   1.459    7.03   0.188   -0.064   0.273      
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

Time-lag meta-regression model. This model tests whether effect sizes change systematically over publication time. A significant slope would indicate temporal trends such as decline or inflation of reported effects.

Time-lag bias: effect size as a function of publication year.

Moderators

Moderator analyses were conducted using multilevel meta-analytic models with robust variance estimation to examine whether effect sizes differed across experimental and biological characteristics. Orchard plots display pooled effects for each moderator level, with study-level clustering and multilevel heterogeneity taken into account.

## 
## Multivariate Meta-Analysis Model (k = 40; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  2.885  1.699     21     no         study_id 
## sigma^2.2  0.000  0.000     28     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 33) = 223.183, p-val < .001
## 
## Number of estimates:   40
## Number of clusters:    21
## Estimates per cluster: 1-5 (mean: 1.90, median: 2)
## 
## Test of Moderators (coefficients 2:7):¹
## F(df1 = 6, df2 = 0) = 0.000, p-val = NA
## 
## Model Results:
## 
##                       estimate     se¹         tval¹     df¹   pval¹   ci.lb¹ 
## intrcpt                  1.622  0.000   4497038.493    3.75   <.001    1.622  
## atp_drugAripiprazole     0.289  0.581         0.498    7.97   0.632   -1.051  
## atp_drugClozapine       -0.503  0.781        -0.644    7.44   0.539   -2.329  
## atp_drugHaloperidol     -4.301  0.955        -4.505    6.36   0.004   -6.606  
## atp_drugOlanzapine       0.710  0.731         0.971    7.14   0.363   -1.012  
## atp_drugRisperidone      0.373  0.583         0.640   15.99   0.531   -0.863  
## atp_drugSulpiride       -1.884  0.881        -2.137    5.98   0.077   -4.043  
##                        ci.ub¹      
## intrcpt                1.622   *** 
## atp_drugAripiprazole   1.629       
## atp_drugClozapine      1.322       
## atp_drugHaloperidol   -1.997    ** 
## atp_drugOlanzapine     2.433       
## atp_drugRisperidone    1.608       
## atp_drugSulpiride      0.275     . 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

## 
## Multivariate Meta-Analysis Model (k = 40; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  1.791  1.338     21     no         study_id 
## sigma^2.2  0.000  0.000     28     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 37) = 328.263, p-val < .001
## 
## Number of estimates:   40
## Number of clusters:    21
## Estimates per cluster: 1-5 (mean: 1.90, median: 2)
## 
## Test of Moderators (coefficients 1:3):¹
## F(df1 = 3, df2 = 2.69) = 2.657, p-val = 0.237
## 
## Model Results:
## 
##                       estimate     se¹   tval¹     df¹   pval¹   ci.lb¹  ci.ub¹ 
## atp_scheduleAcute        0.581  0.346   1.679    4.98   0.154   -0.310   1.471  
## atp_scheduleRepeated     1.154  0.428   2.695   12.93   0.018    0.229   2.080  
## atp_scheduleUnclear      0.298  0.153   1.951       1   0.302   -1.644   2.240  
##                         
## atp_scheduleAcute       
## atp_scheduleRepeated  * 
## atp_scheduleUnclear     
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

## 
## Multivariate Meta-Analysis Model (k = 40; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  1.697  1.303     21     no         study_id 
## sigma^2.2  0.000  0.000     28     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 36) = 349.602, p-val < .001
## 
## Number of estimates:   40
## Number of clusters:    21
## Estimates per cluster: 1-5 (mean: 1.90, median: 2)
## 
## Test of Moderators (coefficients 1:4):¹
## F(df1 = 4, df2 = 0.66) = 0.505, p-val = 0.787
## 
## Model Results:
## 
##                                          estimate     se¹   tval¹     df¹ 
## atp_administration_routeImmersion           0.020  8.793   0.002       1  
## atp_administration_routeIntraperitoneal     0.864  0.348   2.481   11.56  
## atp_administration_routeOral                1.180  0.592   1.994    7.21  
## atp_administration_routeUnclear             0.928  0.660   1.406       1  
##                                           pval¹     ci.lb¹    ci.ub¹    
## atp_administration_routeImmersion        0.999   -111.702   111.743     
## atp_administration_routeIntraperitoneal  0.030      0.102     1.627   * 
## atp_administration_routeOral             0.085     -0.211     2.572   . 
## atp_administration_routeUnclear          0.394     -7.461     9.318     
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

## 
## Multivariate Meta-Analysis Model (k = 40; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  1.530  1.237     21     no         study_id 
## sigma^2.2  0.000  0.000     28     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 37) = 310.765, p-val < .001
## 
## Number of estimates:   40
## Number of clusters:    21
## Estimates per cluster: 1-5 (mean: 1.90, median: 2)
## 
## Test of Moderators (coefficients 1:3):¹
## F(df1 = 3, df2 = 8.63) = 3.528, p-val = 0.064
## 
## Model Results:
## 
##                               estimate     se¹   tval¹    df¹   pval¹   ci.lb¹ 
## nmda_antagonistKetamine          1.291  0.541   2.387   8.95   0.041    0.066  
## nmda_antagonistMK-801            0.131  0.115   1.136   4.99   0.307   -0.166  
## nmda_antagonistPhencyclidine     1.280  0.521   2.459   3.97   0.070   -0.170  
##                               ci.ub¹    
## nmda_antagonistKetamine       2.515   * 
## nmda_antagonistMK-801         0.428     
## nmda_antagonistPhencyclidine  2.730   . 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

## 
## Multivariate Meta-Analysis Model (k = 40; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  1.819  1.349     21     no         study_id 
## sigma^2.2  0.000  0.000     28     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 37) = 326.713, p-val < .001
## 
## Number of estimates:   40
## Number of clusters:    21
## Estimates per cluster: 1-5 (mean: 1.90, median: 2)
## 
## Test of Moderators (coefficients 1:3):¹
## F(df1 = 3, df2 = 1.83) = 1.558, p-val = 0.426
## 
## Model Results:
## 
##                   estimate     se¹   tval¹     df¹   pval¹     ci.lb¹    ci.ub¹ 
## speciesMouse         1.062  0.362   2.932   15.93   0.010      0.294     1.831  
## speciesRat           0.618  0.567   1.091    1.98   0.390     -1.839     3.075  
## speciesZebrafish     0.020  9.163   0.002       1   0.999   -116.410   116.451  
##                      
## speciesMouse      ** 
## speciesRat           
## speciesZebrafish     
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

## 
## Multivariate Meta-Analysis Model (k = 40; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  1.776  1.333     21     no         study_id 
## sigma^2.2  0.000  0.000     28     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 37) = 345.491, p-val < .001
## 
## Number of estimates:   40
## Number of clusters:    21
## Estimates per cluster: 1-5 (mean: 1.90, median: 2)
## 
## Test of Moderators (coefficients 1:3):¹
## F(df1 = 3, df2 = 5.79) = 19.330, p-val = 0.002
## 
## Model Results:
## 
##                                                   estimate     se¹   tval¹ 
## developmental_stage_inductionAdult                   1.119  0.384   2.913  
## developmental_stage_inductionJuvenile/Adolescent     0.521  0.666   0.782  
## developmental_stage_inductionUnclear                 1.293  0.156   8.302  
##                                                     df¹   pval¹   ci.lb¹ 
## developmental_stage_inductionAdult                9.93   0.016    0.262  
## developmental_stage_inductionJuvenile/Adolescent  5.97   0.464   -1.111  
## developmental_stage_inductionUnclear              1.99   0.014    0.620  
##                                                   ci.ub¹    
## developmental_stage_inductionAdult                1.976   * 
## developmental_stage_inductionJuvenile/Adolescent  2.154     
## developmental_stage_inductionUnclear              1.966   * 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

## 
## Multivariate Meta-Analysis Model (k = 40; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  5.257  2.293     21     no         study_id 
## sigma^2.2  0.000  0.000     28     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 27) = 145.522, p-val < .001
## 
## Number of estimates:   40
## Number of clusters:    21
## Estimates per cluster: 1-5 (mean: 1.90, median: 2)
## 
## Test of Moderators (coefficients 2:13):¹
## F(df1 = 12, df2 = 0) = 0.000, p-val = NA
## 
## Model Results:
## 
##                                                  estimate     se¹ 
## intrcpt                                             1.622  0.000  
## atp_nmda_interactionClozapine × Ketamine            0.831  0.663  
## atp_nmda_interactionHaloperidol × Ketamine         -4.871  1.434  
## atp_nmda_interactionOlanzapine × Ketamine           1.213  2.704  
## atp_nmda_interactionRisperidone × Ketamine          1.004  0.707  
## atp_nmda_interactionAripiprazole × MK-801          -0.220  1.257  
## atp_nmda_interactionHaloperidol × MK-801           -7.669  2.212  
## atp_nmda_interactionOlanzapine × MK-801             0.130  1.399  
## atp_nmda_interactionRisperidone × MK-801           -0.237  1.248  
## atp_nmda_interactionSulpiride × MK-801             -4.964  2.238  
## atp_nmda_interactionClozapine × Phencyclidine       0.441  0.433  
## atp_nmda_interactionHaloperidol × Phencyclidine    -1.525  0.409  
## atp_nmda_interactionOlanzapine × Phencyclidine     -1.824  0.000  
##                                                           tval¹    df¹   pval¹ 
## intrcpt                                            3039847.782   5.64   <.001  
## atp_nmda_interactionClozapine × Ketamine                 1.254   4.27   0.274  
## atp_nmda_interactionHaloperidol × Ketamine              -3.397    4.2   0.025  
## atp_nmda_interactionOlanzapine × Ketamine                0.449      1   0.732  
## atp_nmda_interactionRisperidone × Ketamine               1.420   6.83   0.200  
## atp_nmda_interactionAripiprazole × MK-801               -0.175    4.8   0.868  
## atp_nmda_interactionHaloperidol × MK-801                -3.468   3.27   0.035  
## atp_nmda_interactionOlanzapine × MK-801                  0.093   4.75   0.930  
## atp_nmda_interactionRisperidone × MK-801                -0.190   4.73   0.857  
## atp_nmda_interactionSulpiride × MK-801                  -2.218   3.37   0.103  
## atp_nmda_interactionClozapine × Phencyclidine            1.018      3   0.384  
## atp_nmda_interactionHaloperidol × Phencyclidine         -3.729   2.97   0.034  
## atp_nmda_interactionOlanzapine × Phencyclidine   -15188416.334   5.56   <.001  
##                                                    ci.lb¹   ci.ub¹      
## intrcpt                                            1.622    1.622   *** 
## atp_nmda_interactionClozapine × Ketamine          -0.965    2.628       
## atp_nmda_interactionHaloperidol × Ketamine        -8.780   -0.962     * 
## atp_nmda_interactionOlanzapine × Ketamine        -33.149   35.575       
## atp_nmda_interactionRisperidone × Ketamine        -0.676    2.685       
## atp_nmda_interactionAripiprazole × MK-801         -3.493    3.053       
## atp_nmda_interactionHaloperidol × MK-801         -14.395   -0.943     * 
## atp_nmda_interactionOlanzapine × MK-801           -3.522    3.783       
## atp_nmda_interactionRisperidone × MK-801          -3.501    3.026       
## atp_nmda_interactionSulpiride × MK-801           -11.667    1.739       
## atp_nmda_interactionClozapine × Phencyclidine     -0.937    1.818       
## atp_nmda_interactionHaloperidol × Phencyclidine   -2.835   -0.216     * 
## atp_nmda_interactionOlanzapine × Phencyclidine    -1.824   -1.824   *** 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

Meta-regression

Cumulative exposure

## 
## Multivariate Meta-Analysis Model (k = 34; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  1.833  1.354     19     no         study_id 
## sigma^2.2  0.000  0.000     26     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 32) = 292.990, p-val < .001
## 
## Number of estimates:   34
## Number of clusters:    19
## Estimates per cluster: 0-5 (mean: 1.62, median: 1)
## 
## Test of Moderators (coefficient 2):¹
## F(df1 = 1, df2 = 1.14) = 0.003, p-val = 0.963
## 
## Model Results:
## 
##                          estimate     se¹   tval¹     df¹   pval¹   ci.lb¹ 
## intrcpt                     1.028  0.334   3.076   17.61   0.007    0.325  
## atp_cumulative_exposure     0.000  0.002   0.057    1.14   0.963   -0.022  
##                          ci.ub¹     
## intrcpt                  1.730   ** 
## atp_cumulative_exposure  0.022      
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

Meta-regression of cumulative exposure versus effect size. The regression coefficient indicates whether increasing cumulative exposure is associated with changes in effect size, suggesting a potential dose–response relationship.

Log-transformed cumulative exposure

## 
## Multivariate Meta-Analysis Model (k = 34; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  1.857  1.363     19     no         study_id 
## sigma^2.2  0.000  0.000     26     no  study_id/exp_id 
## 
## Test for Residual Heterogeneity:
## QE(df = 32) = 291.254, p-val < .001
## 
## Number of estimates:   34
## Number of clusters:    19
## Estimates per cluster: 0-5 (mean: 1.62, median: 1)
## 
## Test of Moderators (coefficient 2):¹
## F(df1 = 1, df2 = 1.58) = 0.334, p-val = 0.635
## 
## Model Results:
## 
##                              estimate     se¹    tval¹     df¹   pval¹   ci.lb¹ 
## intrcpt                         1.145  0.387    2.958   13.06   0.011    0.309  
## log_atp_cumulative_exposure    -0.133  0.230   -0.578    1.58   0.635   -1.424  
##                              ci.ub¹    
## intrcpt                      1.981   * 
## log_atp_cumulative_exposure  1.158     
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t/F-tests and confidence intervals, df: Satterthwaite approx)

Meta-regression of log-transformed cumulative exposure. The log-transformed model evaluates potential non-linear exposure–effect relationships and the robustness of the association.

Sensitivity analyses

Rho sensitivity

##   rho  estimate             ci
## 1 0.0 1.1634533  [0.66, 1.667]
## 2 0.3 1.0519299  [0.52, 1.584]
## 3 0.5 0.9397334 [0.348, 1.531]
## 4 0.8 0.5965782 [-0.237, 1.43]

Sensitivity of the overall effect to within-study correlation (rho). This analysis evaluates the robustness of the pooled effect size to assumptions about the correlation between multiple effect sizes within the same experiment. Each row reports the overall effect estimate (Hedges’ g) and 95% confidence interval obtained under a different assumed value of rho. Stability of estimates across rho values indicates robustness to within-study dependency assumptions.

Leave-one-study-out

##      left_out_study  estimate     ci_lb    ci_ub
## 1       araujo_2016 0.8814916 0.2738560 1.489127
## 2       araujo_2021 0.8398390 0.2546932 1.424985
## 3      ben-azu_2018 1.0718716 0.5300165 1.613727
## 4     ben-azu_2018b 0.7676359 0.2557572 1.279514
## 5     ben-azu_2018c 0.9975420 0.3814165 1.613667
## 6       deiana_2015 1.0140873 0.4062520 1.621923
## 7       gil-ad_2014 0.9948057 0.3837137 1.605898
## 8        jeong_2022 0.9750287 0.3543764 1.595681
## 9           ju_2020 0.9747979 0.3544917 1.595104
## 10         koo_2019 0.9752219 0.3546167 1.595827
## 11       monte_2013 0.9288168 0.3038531 1.553780
## 12   nikiforuk_2016 0.9099172 0.2930953 1.526739
## 13      oshodi_2021 0.8712722 0.2660169 1.476528
## 14      sanavi_2022 0.9519570 0.3269470 1.576967
## 15       seibt_2011 0.9907319 0.3722333 1.609230
## 16      tadmor_2018 0.8964300 0.2812023 1.511658
## 17        tran_2016 0.9504091 0.3273866 1.573432
## 18       tran_2016b 0.9352433 0.3136109 1.556876
## 19        tran_2018 0.8474022 0.2630749 1.431730
## 20 vasconcelos_2015 0.9848004 0.3695580 1.600043
## 21         xue_2017 0.9761117 0.3562122 1.596011

Leave-one-study-out analysis. Each row reports the pooled effect size (Hedges’ g) and 95% confidence interval obtained after excluding one study at a time from the meta-analysis. This analysis evaluates the influence of individual studies on the overall estimate; substantial changes after removal of a study would indicate disproportionate influence.

Excluding high risk of bias

## 
## Multivariate Meta-Analysis Model (k = 11; method: REML)
## 
## Variance Components:
## 
##            estim   sqrt  nlvls  fixed           factor 
## sigma^2.1  0.405  0.636      8     no         study_id 
## sigma^2.2  0.000  0.000      9     no  study_id/exp_id 
## 
## Test for Heterogeneity:
## Q(df = 10) = 44.550, p-val < .001
## 
## Number of estimates:   11
## Number of clusters:    8
## Estimates per cluster: 0-2 (mean: 0.52, median: 0)
## 
## Model Results:
## 
## estimate     se¹   tval¹    df¹   pval¹  ci.lb¹  ci.ub¹    
##    0.790  0.298   2.654   6.69   0.034   0.080   1.501   * 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## 1) results based on cluster-robust inference (var-cov estimator: CR2,
##    approx t-test and confidence interval, df: Satterthwaite approx)

Overall effect excluding high risk-of-bias studies. This sensitivity analysis re-estimates the overall meta-analytic effect after excluding studies classified as having high risk of bias. The purpose of this analysis is to assess whether the pooled effect estimate is robust to the exclusion of potentially biased evidence.

Annex: Individual effect sizes included in the meta-analysis

Calculated effect sizes

## 
##               study effect_id   species nmda_antagonist     atp_drug hedges_g 
## 1      Ben-Azu 2018         5     Mouse        Ketamine  Risperidone    6.972 
## 2      Ben-Azu 2018         4     Mouse        Ketamine  Haloperidol   -1.960 
## 3    Nikiforuk 2016        25       Rat        Ketamine  Amisulpride    1.622 
## 4    Ben-Azu 2018 b       110     Mouse        Ketamine  Risperidone    3.836 
## 5    Ben-Azu 2018 b       111     Mouse        Ketamine  Risperidone    6.249 
## 6       Deiana 2015       118       Rat          MK-801  Risperidone   -0.301 
## 7       Deiana 2015       115       Rat          MK-801 Aripiprazole   -0.448 
## 8       Deiana 2015       119       Rat          MK-801   Olanzapine   -0.175 
## 9       Deiana 2015       116       Rat          MK-801 Aripiprazole   -0.332 
## 10      Deiana 2015       117       Rat          MK-801 Aripiprazole   -0.437 
## 11       Seibt 2011       155 Zebrafish          MK-801  Haloperidol    0.519 
## 12       Seibt 2011       156 Zebrafish          MK-801    Sulpiride    3.480 
## 13       Seibt 2011       157 Zebrafish          MK-801   Olanzapine   11.674 
## 14      Tadmor 2018       161     Mouse   Phencyclidine    Clozapine    1.830 
## 15        Tran 2018       162     Mouse   Phencyclidine    Clozapine    3.206 
## 16   Ben-Azu 2018 c       167     Mouse        Ketamine  Risperidone    5.500 
## 17   Ben-Azu 2018 c       166     Mouse        Ketamine  Haloperidol    0.157 
## 18   Ben-Azu 2018 c       169     Mouse        Ketamine  Risperidone    5.285 
## 19   Ben-Azu 2018 c       168     Mouse        Ketamine  Haloperidol   -0.108 
## 20      Araujo 2021       170     Mouse        Ketamine   Olanzapine    2.246 
## 21      Araujo 2021       171     Mouse        Ketamine   Olanzapine    3.874 
## 22          Ju 2020       174     Mouse          MK-801  Risperidone    0.086 
## 23          Ju 2020       175     Mouse          MK-801 Aripiprazole    0.333 
## 24          Ju 2020       173     Mouse          MK-801   Olanzapine    0.489 
## 25         Koo 2019       216     Mouse          MK-801 Aripiprazole    0.298 
## 26      Oshodi 2021       229     Mouse        Ketamine  Risperidone    2.142 
## 27      Oshodi 2021       230     Mouse        Ketamine  Risperidone    2.394 
## 28      Araujo 2016       270     Mouse        Ketamine  Risperidone    2.178 
## 29      Sanavi 2022       272       Rat        Ketamine  Risperidone    0.778 
## 30      Sanavi 2022       271       Rat        Ketamine    Clozapine    0.746 
## 31      Gil-Ad 2014       298     Mouse   Phencyclidine   Olanzapine   -0.203 
## 32        Tran 2016       337     Mouse   Phencyclidine  Haloperidol    0.234 
## 33        Tran 2016       336     Mouse   Phencyclidine    Clozapine    2.208 
## 34      Tran 2016 b       338     Mouse   Phencyclidine    Clozapine    1.086 
## 35       Jeong 2022       414     Mouse          MK-801 Aripiprazole    0.302 
## 36       Monte 2013       415     Mouse        Ketamine  Risperidone    1.082 
## 37       Monte 2013       416     Mouse        Ketamine  Risperidone    1.309 
## 38 Vasconcelos 2015       472     Mouse        Ketamine    Clozapine    2.476 
## 39 Vasconcelos 2015       473     Mouse        Ketamine    Clozapine   -0.904 
## 40         Xue 2017       478     Mouse          MK-801  Risperidone    0.270 
##     ci_lb  ci_ub 
## 1   4.185  9.759 
## 2  -3.235 -0.686 
## 3   0.317  2.926 
## 4   2.070  5.601 
## 5   3.708  8.790 
## 6  -0.990  0.388 
## 7  -1.143  0.247 
## 8  -1.012  0.662 
## 9  -0.844  0.179 
## 10 -0.932  0.057 
## 11 -0.331  1.369 
## 12  2.155  4.805 
## 13  8.125 15.223 
## 14  0.786  2.874 
## 15  1.495  4.916 
## 16  3.357  7.642 
## 17 -0.825  1.138 
## 18  3.208  7.362 
## 19 -1.089  0.872 
## 20  0.994  3.497 
## 21  2.212  5.535 
## 22 -1.101  1.274 
## 23 -0.862  1.528 
## 24 -0.715  1.693 
## 25 -0.631  1.227 
## 26  0.828  3.457 
## 27  1.022  3.767 
## 28  0.941  3.415 
## 29 -0.180  1.736 
## 30 -0.209  1.702 
## 31 -1.445  1.040 
## 32 -0.902  1.369 
## 33  0.772  3.643 
## 34 -0.126  2.298 
## 35 -0.627  1.231 
## 36  0.033  2.131 
## 37  0.229  2.389 
## 38  0.829  4.124 
## 39 -2.205  0.397 
## 40 -0.714  1.254

Individual effect sizes included in the meta-analysis. This table lists all calculated Hedges’ g values and corresponding confidence intervals used in the analyses.

Session info

## R version 4.3.1 (2023-06-16 ucrt)
## Platform: x86_64-w64-mingw32/x64 (64-bit)
## Running under: Windows 11 x64 (build 26100)
## 
## Matrix products: default
## 
## 
## locale:
## [1] LC_COLLATE=Portuguese_Brazil.utf8  LC_CTYPE=Portuguese_Brazil.utf8   
## [3] LC_MONETARY=Portuguese_Brazil.utf8 LC_NUMERIC=C                      
## [5] LC_TIME=Portuguese_Brazil.utf8    
## 
## time zone: America/Sao_Paulo
## tzcode source: internal
## 
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## other attached packages:
##  [1] RColorBrewer_1.1-3  scales_1.4.0        stringr_1.5.1      
##  [4] forcats_1.0.1       ggalluvial_0.12.5   tidyr_1.3.1        
##  [7] ggplot2_4.0.0       orchaRd_2.1.3       clubSandwich_0.6.1 
## [10] metafor_4.8-0       numDeriv_2016.8-1.1 metadat_1.2-0      
## [13] Matrix_1.6-5        dplyr_1.1.4         readxl_1.4.5       
## 
## loaded via a namespace (and not attached):
##  [1] gtable_0.3.6       beeswarm_0.4.0     xfun_0.52          bslib_0.9.0       
##  [5] lattice_0.22-6     mathjaxr_1.6-0     vctrs_0.6.5        tools_4.3.1       
##  [9] generics_0.1.4     sandwich_3.1-1     tibble_3.2.1       pkgconfig_2.0.3   
## [13] S7_0.2.0           lifecycle_1.0.4    compiler_4.3.1     farver_2.1.2      
## [17] textshaping_1.0.0  prettydoc_0.4.1    codetools_0.2-20   vipor_0.4.7       
## [21] htmltools_0.5.8.1  sass_0.4.9         yaml_2.3.10        pillar_1.11.1     
## [25] jquerylib_0.1.4    MASS_7.3-60.0.1    cachem_1.1.0       multcomp_1.4-28   
## [29] nlme_3.1-164       tidyselect_1.2.1   digest_0.6.35      mvtnorm_1.3-3     
## [33] stringi_1.8.7      purrr_1.0.2        labeling_0.4.3     splines_4.3.1     
## [37] latex2exp_0.9.6    fastmap_1.2.0      grid_4.3.1         cli_3.6.2         
## [41] magrittr_2.0.3     survival_3.5-8     TH.data_1.1-4      withr_3.0.2       
## [45] ggbeeswarm_0.7.2   estimability_1.5.1 rmarkdown_2.30     emmeans_1.11.2-8  
## [49] cellranger_1.1.0   ragg_1.3.3         zoo_1.8-13         evaluate_1.0.5    
## [53] knitr_1.50         rlang_1.1.5        xtable_1.8-4       glue_1.8.0        
## [57] rstudioapi_0.17.1  jsonlite_2.0.0     R6_2.6.1           systemfonts_1.2.2